# forck.live — 23–29 September 2024 > AI lab announcements, news reports and podcasts. forck.live follows what companies announce about their own products. Press may corroborate an official story; it never leads and never appears as an item of its own. A missing date stays missing. A title is the publisher's words — where a company does not publish in English, an English rendering is listed separately as `Title in English` and named as this site's translation, never as theirs. This file is one issue of the weekly archive: every confirmed first-party announcement of ISO week 2024-W39, 23–29 September 2024, newest first. The window never moves, so this URL always names the same seven days — but a post published inside them and detected later is added when it arrives, so a closed week can still gain entries. Every entry is one first-party announcement, and carries the same eight fields in this order before any others: Company · Announced · Category · Coverage · Announcement · Group · Source · Record `Source` is the company's own page the announcement was traced to and `Record` is this site's page for it. Both are safe to cite. `Coverage` is a number of outlets, or `not counted` — which means this record reached no count of its own, never that nobody wrote about it. `Category` is this tracker's own analysis, and is authoritative wherever it is stated. `not stated` means either that nothing has analysed the post yet or that the analysis placed it outside this taxonomy; neither is a verdict about the announcement. `Announcement` is whether the post is news of something the company did, rather than a how-to, a customer story, a changelog listing or a staffing note. `Group` is which of this record's groups the entry falls in, carried on the entry itself so it holds where the groups are headings and where they are not. One of: `models`, `covered`, `announcements`, `routine`. Anything after `Record` is held detail that only some announcements carry. - This week's page: https://forck.live/week/2024-W39 - The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md - Catalogue: https://forck.live/llms.txt ## Announcements ### Vercel — Accelerating developer velocity and creating high-impact web teams - Company: Vercel (vercel.com) - Announced: 2024-09-27T13:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://vercel.com/blog/accelerating-developer-velocity-and-creating-high-impact-web-teams - Record: https://forck.live/items/20945-accelerating-developer-velocity-and-creating-high-impact-web-teams - Subject: v0 / AI SDK / AI Gateway Vercel's Frontend Cloud, powered by framework-defined infrastructure, handles cloud infrastructure so teams can focus on coding user interfaces and logic. A Forrester Total Economic Impact report found that developers spend 90% less time managing frontend infrastructure and 80% less time building and deploying code on Vercel, and release four times more major website enhancements, improving website performance by up to 90%. The report also states that higher customer conversion rates on Vercel generate $2.6 million in incremental profits and higher website traffic generates $7.7 million in incremental profits. ### Harvey — BigLaw Bench Workflows: SPA Deal Points - Company: Harvey (harvey.ai) - Announced: 2024-09-26T23:37:00+00:00 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://www.harvey.ai/blog/biglaw-bench-workflows-spa-deal-points - Record: https://forck.live/items/17724-biglaw-bench-workflows-spa-deal-points - Subject: Harvey Harvey is expanding its publicly-released BigLaw Bench data to include workflow datasets for evaluating legal workflow agents, starting with extracting deal points from Share Purchase Agreements (SPAs). Harvey's SPA agents achieve 98.47% accuracy on this task, compared to 66.04% (GPT-4o) and 72.27% (Gemini) for general-purpose foundation models. The company plans to release more workflow datasets and incorporate these workflows into its Assistant and Vault products. ### OpenAI — Upgrading the Moderation API with our new multimodal moderation model - Company: OpenAI (openai.com) - Announced: 2024-09-26T10:00:00+00:00 - Category: api-release - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/upgrading-the-moderation-api-with-our-new-multimodal-moderation-model - Record: https://forck.live/items/702-upgrading-the-moderation-api-with-our-new-multimodal-moderation-model - Subject: GPT / ChatGPT / API OpenAI announces a new moderation model based on GPT-4o, improving accuracy in detecting harmful text and images, and making it available via the Moderation API for developers to build better moderation systems. ### OpenAI — Minnesota’s Enterprise Translation Office uses ChatGPT to bridge language gaps - Company: OpenAI (openai.com) - Announced: 2024-09-26T07:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/state-of-minnesota - Record: https://forck.live/items/703-minnesota-s-enterprise-translation-office-uses-chatgpt-to-bridge-language-gaps - Subject: GPT / ChatGPT / API Minnesota's Enterprise Translation Office uses ChatGPT to bridge language gaps. ### OpenAI — OpenAI and GEDI partner for Italian news content - Company: OpenAI (openai.com) - Announced: 2024-09-26T04:30:00+00:00 - Category: partnership-acquisition - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/gedi - Record: https://forck.live/items/704-openai-and-gedi-partner-for-italian-news-content - Subject: GPT / ChatGPT / API OpenAI and GEDI have announced a strategic partnership to bring Italian-language news content to ChatGPT. ### LG AI Research — [ACL 2024] Developing a Large Language Model with Cross-Cultural Considerations - Company: LG AI Research (lgresearch.ai) - Announced: 2024-09-26 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://www.lgresearch.ai/blog/view?seq=487 - Record: https://forck.live/items/4826-acl-2024-developing-a-large-language-model-with-cross-cultural-considerations - Subject: EXAONE LG AI Research discusses cross-cultural considerations in LLMs, referencing their EXAONE 3.0 7.8B model and research presented at the C3NLP workshop at ACL 2024. ### Hugging Face — Llama can now see and run on your device - welcome Llama 3.2 - Company: Hugging Face (huggingface.co) - Announced: 2024-09-25 - Category: open-weight-release - Coverage: not counted - Announcement: yes - Group: models - Source: https://huggingface.co/blog/llama32 - Record: https://forck.live/items/1825-llama-can-now-see-and-run-on-your-device-welcome-llama-3-2 - Subject: Platform Meta and Hugging Face release Llama 3.2, including multimodal vision models (11B and 90B) and small text-only models (1B and 3B), all open-weight, with Llama Guard 3 for safety. ### Vercel — Preventing infrastructure abuse with Vercel Firewall - Company: Vercel (vercel.com) - Announced: 2024-09-24T13:00:00+00:00 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://vercel.com/blog/preventing-infrastructure-abuse-with-vercel-firewall - Record: https://forck.live/items/20946-preventing-infrastructure-abuse-with-vercel-firewall - Subject: v0 / AI SDK / AI Gateway Vercel describes its Firewall's multi-layered DDoS protection, which blocks around 1 billion suspicious TCP connections weekly and recently mitigated a 1.37 Tbps attack with no downtime. The Firewall provides automated defense across all user tiers, integrating Layer 4 and Layer 7 protections, and includes a customizable WAF for application-level threats. ### OpenAI — Introducing Verdi, an AI dev platform powered by GPT-4o - Company: OpenAI (openai.com) - Announced: 2024-09-24T07:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://openai.com/index/mercado-libre - Record: https://forck.live/items/705-introducing-verdi-an-ai-dev-platform-powered-by-gpt-4o - Subject: GPT / ChatGPT / API The source text indicates that Mercado Libre is introducing an AI developer platform called Verdi that is powered by GPT-4o. ### LG AI Research — [ACL 2024] Research Trends in LLM Evaluation Methods for Faithfulness and LLM Efficiency - Company: LG AI Research (lgresearch.ai) - Announced: 2024-09-24 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/blog/view?seq=485 - Record: https://forck.live/items/4827-acl-2024-research-trends-in-llm-evaluation-methods-for-faithfulness-and-llm - Subject: EXAONE This is a blog post from LG AI Research summarizing research trends presented at ACL 2024, focusing on LLM evaluation methods for faithfulness and LLM efficiency. It discusses various papers and topics but does not announce any new product, model, or update from LG AI Research. ### Harvey — BigLaw Bench – Sources - Company: Harvey (harvey.ai) - Announced: 2024-09-23T21:41:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.harvey.ai/blog/biglaw-bench-sources - Record: https://forck.live/items/17725-biglaw-bench-sources - Subject: Harvey Harvey launched BigLaw Bench, a public benchmark for evaluating large language models on complex legal tasks, with a focus on source referencing accuracy. The benchmark introduces a 'source score' to measure how well model answers connect factual statements to source documents. Harvey reports that its systems substantially outperform other LLMs on this metric. ### Lovable — Branch Switching, Hot Projects, and Reliability Improvements - Company: Lovable (lovable.dev) - Announced: 2024-09-23T11:04:00+00:00 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://lovable.dev/blog/2025-01-13-branch-switching-hot-projects-and-reliability-improvements - Record: https://forck.live/items/12206-branch-switching-hot-projects-and-reliability-improvements - Subject: Lovable / AI app builder GPT Engineer added branch switching, a Hot Projects tab, network error detection, RAG-based context management, and automatic placeholder completion to improve reliability and support larger projects. ### Supabase — Local-first Realtime Apps with Expo and Legend-State - Company: Supabase (supabase.com) - Announced: 2024-09-23 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://supabase.com/blog/local-first-expo-legend-state - Record: https://forck.live/items/21397-local-first-realtime-apps-with-expo-and-legend-state - Subject: Supabase Supabase published a tutorial on building local-first realtime applications using Expo and Legend-State, covering setup, configuration, data syncing, and realtime updates with Supabase. ### LG AI Research — LG AI Research Takes the First Step in Leading Global AI Ethics Standards - Company: LG AI Research (lgresearch.ai) - Announced: 2024-09-23 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/news/view?seq=480 - Record: https://forck.live/items/11019-lg-ai-research-takes-the-first-step-in-leading-global-ai-ethics-standards - Subject: EXAONE LG AI Research has become the first organization in Korea to be designated an IEEE CertifAIEd Assessment Collaborator, partnering with IEEE-SA to lead the development of international AI ethics standards. The company plans to prioritize the IEEE CertifAIEd certification program for AI products and services of LG affiliates and global partners, and will operate AI ethics education programs. LG AI Research has also joined the UNESCO Business Council and the AI Election Compact, and will present at a UN Global Compact AI Safety session. ### Suno — Introducing Sunoversity - Company: Suno (suno.com) - Announced: 2024-09-23 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://suno.com/blog/suno-students - Record: https://forck.live/items/10342-introducing-sunoversity - Subject: Music / Studio Suno introduced Sunoversity, a campus ambassador program for undergraduate and master's students, offering perks such as free Suno Premier, early access to features, and collaboration with Suno's founders. Applications close October 13, 2024. ### LG AI Research — LG AI Research Takes the First Step in Leading Global AI Ethics Standards - Company: LG AI Research (lgresearch.ai) - Announced: 2024-09-23 - Category: partnership-acquisition - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://www.lgresearch.ai/blog/view?seq=480 - Record: https://forck.live/items/4828-lg-ai-research-takes-the-first-step-in-leading-global-ai-ethics-standards - Subject: EXAONE LG AI Research has been designated as the first Korean organization to become an IEEE CertifAIEd Assessment Collaborator, aiming to lead international AI ethics standards. ### Hugging Face — Exploring the Daily Papers Page on Hugging Face - Company: Hugging Face (huggingface.co) - Announced: 2024-09-23 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://huggingface.co/blog/daily-papers - Record: https://forck.live/items/1827-exploring-the-daily-papers-page-on-hugging-face - Subject: Platform The features of the Hugging Face Daily Papers page, including claiming papers, submitting papers, chatting with authors, accessing related resources, upvoting, receiving similar paper recommendations, multilingual comments with translation, subscribing to daily updates, and integration with arXiv. ### Hugging Face — FineVideo: behind the scenes - Company: Hugging Face (huggingface.co) - Announced: 2024-09-23 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://huggingface.co/blog/fine-video - Record: https://forck.live/items/1826-finevideo-behind-the-scenes - Subject: Platform Hugging Face announces FineVideo, a dataset of 43k videos (3.4k hours) with rich annotations, built from YouTube-Commons to advance open-source video AI. The post details the pipeline: filtering English videos, downloading 1.8M videos, applying word density and visual dynamism filters, categorizing with a custom taxonomy, and annotating with Gemini 1.5 Pro and GPT-4o.